• About
  • Advanced Search
  • Browse Proceedings
  • Access Policy
  • Sponsor
  • Contact
  • Gallery Index
  • Vision-LiDAR Based Relative Navigation Architecture for Autonomous Rendezvous with Tumbling Space Debris

    Paper ID

    110586

    authors

    • Shabadini Sampath
    • Feng Jinglang

    company

    University of Strathclyde / Mechanical and Aerospace Engineering; University of Strathclyde

    country

    United Kingdom

    year

    2026

    abstract

    As space traffic density increases, autonomous rendezvous with non-cooperative and tumbling objects has become a critical requirement for debris removal and on-orbit servicing missions. Unlike cooperative spacecraft equipped with GNSS receivers or navigation beacons, unprepared debris targets provide no onboard navigation support, requiring fully autonomous, space-based relative navigation strategies capable of operating under uncertain and dynamically evolving conditions. This paper presents an integrated vision-LiDAR relative navigation architecture for real-time pose determination and synchronisation between a servicing spacecraft and a tumbling debris object. The proposed framework combines Image-Based Visual Servoing (IBVS) with LiDAR-assisted depth sensing to estimate relative position, attitude, and angular motion during close-proximity operations. Vision-based feature tracking enables continuous estimation of relative orientation and motion trends, while LiDAR measurements improve range observability and robustness against scale ambiguity and lighting variability. The fusion of these sensing modalities enhances navigation reliability during approach and capture preparation. The architecture is designed to operate without reliance on external navigation infrastructure, enabling autonomy in Earth orbit and potentially cislunar environments. Relative state estimates are generated to guide synchronisation between the spacecraft and the rotating target prior to capture. To address uncertainties arising from unknown debris inertia properties, varying angular velocities, and unmodeled disturbances, a neural network based sliding mode controller adapts to dynamic variations while maintaining stable tracking performance. This control component operates on top of the navigation outputs, ensuring accurate convergence without compromising estimation integrity. Simulation studies conducted in MATLAB evaluate the framework under realistic tumbling scenarios and sensor uncertainties. Results demonstrate robust relative pose convergence, improved tracking accuracy, and reduced synchronisation time compared to conventional estimation-only approaches. The integrated sensing architecture maintains stable performance even under high angular velocity variations and partial feature visibility conditions. The proposed framework contributes to the advancement of space-based autonomous navigation capabilities for non-cooperative targets. By integrating complementary sensing modalities into a unified relative navigation architecture, this work supports the development of reliable rendezvous and capture strategies essential for sustainable space operations and future servicing missions.

    STATUS: waiting for script files to load